Which statistic is commonly used to estimate internal consistency and may be low if item variance is high relative to total test variance?

Which statistic is commonly used to estimate internal consistency and may be low if item variance is high relative to total test variance?

• A. Cronbach's alpha
• B. Spearman-Brown coefficient
• C. Kappa
• D. Pearson r

Answer: (A)

Internal consistency is about whether all the items on a scale hang together and measure the same underlying construct. The statistic most commonly used for this is Cronbach's alpha. It combines how well items relate to each other (inter-item correlations) with how many items there are. When item variance is large relative to the total test variance, each item behaves more differently from the overall score, so the average inter-item correlation drops and Cronbach's alpha decreases. In other words, high item-to-total variance can make the items seem less cohesive, lowering the internal consistency estimate. Cronbach's alpha tends to rise with stronger inter-item correlations and with more items, so interpretation should consider both factors. The other options serve different purposes: the Spearman-Brown coefficient is used to estimate reliability when changing the test length or in split-half contexts; Kappa measures agreement for categorical ratings; Pearson correlation assesses the relationship between two variables rather than the overall internal consistency of a multi-item scale.

Internal consistency is about whether all the items on a scale hang together and measure the same underlying construct. The statistic most commonly used for this is Cronbach's alpha. It combines how well items relate to each other (inter-item correlations) with how many items there are. When item variance is large relative to the total test variance, each item behaves more differently from the overall score, so the average inter-item correlation drops and Cronbach's alpha decreases. In other words, high item-to-total variance can make the items seem less cohesive, lowering the internal consistency estimate.

Cronbach's alpha tends to rise with stronger inter-item correlations and with more items, so interpretation should consider both factors. The other options serve different purposes: the Spearman-Brown coefficient is used to estimate reliability when changing the test length or in split-half contexts; Kappa measures agreement for categorical ratings; Pearson correlation assesses the relationship between two variables rather than the overall internal consistency of a multi-item scale.